Meta-Analysis of the Heterogeneity in Association of DRD4 7-Repeat Allele and AD/HD: Stronger Association with AD/HD Combined Type Taylor F

RESEARCH ARTICLE

Meta-Analysis of the Heterogeneity in Association of DRD4 7-Repeat Allele and AD/HD: Stronger Association With AD/HD Combined Type Taylor F. Smith* Department of Psychology, University of North Carolina at Greensboro, Greensboro, North Carolina

The purpose of this meta-analysis was to examine whether association studies between attention deficit/hyperactivity dis order (AD/HD) and the dopamine receptor 4 gene 7-repeat (DRD4 7R) allele vary systematically based on study character istics. A total of 27 empirical studies with 28 distinct samples using either case–control or family-based association analyses were included. Consistent with previous meta-analytic work [Gizer et al. (2009), Hum Genet 126:51–90], the DRD4 7R allele was associated with AD/HD across studies (OR ¼ 1.33; 95% CI ¼ 1.16–1.53, z ¼ 4.04, P ¼ 0.00005) and there was significant systematic variability among studies (Q ¼ 54.24; P ¼ 0.001; tion and hyperactive-impulsive symptoms; AD/HD predominantly I2 ¼ 50.22). To account for the variability among studies, sample inattentive type (AD/HD-I) if they display excessive inattention and study level covariates were examined. No differences in only; and AD/HD predominantly hyperactive-impulsive type overall effect size emerged between family-based and case– (AD/HD-HI) if they show excessive hyperactive-impulsive symp control studies. However, the risk allele frequency in the control toms only. AD/HD-I is the most prevalent subtype in community population accounted for a significant portion of the variance samples [e.g., Gaub and Carlson, 1997; DuPaul et al., 1998]. In in overall effect size within case–control studies. In addition, contrast, AD/HD-C tends to be the most prevalent subtype in evidence for the association between the DRD4 7R allele and clinical populations, outnumbering AD/HD-I by 2:1 and AD/HD distinct AD/HD subtypes emerged across family-based and HI by 3:1 [Lahey et al., 1994]. Given that AD/HD-HI is the least case–control studies. The proportion of AD/HD, combined type prevalent subtype and lacks temporal stability [Lahey et al., 2005], individuals within the AD/HD sample was associated with a this subtype will not be further discussed in detail. significant increase in the magnitude of association between the Although the specific etiology is not completely understood, DRD4 7R allele and AD/HD. Conversely, an increase in the several factors associated with the development of AD/HD have proportion of AD/HD, predominantly inattentive type individ been identified. Evidence from family, twin, and adoption studies uals within the AD/HD sample was associated with a decrease in suggest that genetic factors substantially contribute to the devel study effect size. Implications regarding AD/HD etiological and opment of AD/HD [heritability estimate ¼ 0.76; Faraone et al., phenotypic heterogeneity are discussed. 2005]. Molecular genetics research has attempted to identify genes that increase susceptibility for the disorder. Consistent with the Key words: attention; hyperactivity; dopamine; genetics; meta dopamine deficit hypothesis of AD/HD etiology [Levy, 1991], genes regression associated with the dopamine system (e.g., DAT1 and DRD4) have been major foci of study. The dopamine receptor 4 gene 7 repeat INTRODUCTION (DRD4 7R) allele has been widely studied and is one of the most strongly associated alleles with AD/HD [OR ¼ 1.33; Gizer et al., Attention deficit/hyperactivity disorder (AD/HD) is characterized 2009]. by persistent, pervasive and developmentally inappropriate levels of inattention, hyperactivity–impulsivity, or both that lead to clini *Correspondence to: cally significant impairment. The AD/HD phenotype is highly Taylor F. Smith, AD/HD Clinic at UNCG, University of North Carolina at heterogeneous and there has been much debate on how to appro Greensboro, 1100 W. Market St. 3rd Floor, P.O. Box 26170, Greensboro, priately reduce phenotypic heterogeneity by creating diagnostic NC 27402. E-mail: [email protected] subtypes or separate disorders [e.g., Milich et al., 2001]. According to the DSM-IV, individuals are sub-grouped into AD/HD com bined type (AD/HD-C) if they exhibit high levels of both inatten The DRD4 gene is a 48-bp VNTR on exon 3, located on defiant disorder (ODD) and found that the DRD4 7R allele tends to chromosome 11p15.5 [Gelernter et al., 1992; Petronis et al., be more strongly associated with AD/HD when comorbid with 1993]. Allele variants produce structural differences in the 3rd ODD [Holmes et al., 2002; Kirley et al., 2004]. intracellular loop of the D4 receptor, which belongs to the D2 Such inconsistent findings may be related to the small sample receptor family, and couples to pre- and post-synaptic G-protein sizes and limited stability of the DSM-IV AD/HD subtypes overtime effectors. D4 receptors are found at high densities in the frontal [Lahey et al., 2005]; thus, this question may be better addressed in cortex, amygdala, hippocampus, hypothalamus, and mesencepha the context of a mega-sample or meta-analysis [Faraone, 2008]. For lon [Van Tol et al., 1991; O’Malley et al., 1992], and the 7R allele may example, by combining 14 independent samples, Lowe et al. [2004] be associated with a sub-sensitive post-synaptic D4 receptor or demonstrated that the DRD5 148 bp allele was associated with AD/ hypofunctioning of mesolimbic and mesocortical dopamine HD-C and AD/HD-I but not AD/HD-HI. Similarly, I.D. Waldman branches [Missale et al., 1998]. Ten alleles (2R-11R) have been (personal communication, September 8, 2009) used summary identified in the global population [Seeman and Van Tol, 1994] statistics between studies and found evidence for a stronger rela with the 2R, 4R and 7R alleles being the most common variants. tionship between DAT1 and 5HTTLPR risk alleles and AD/HD-C DRD4 allele frequency varies substantially between ethnic compared to AD/HD-I. groups [Chang et al., 1996]. Generally, the 7R allele is Study methodology. Others have taken a between-studies ap relatively most prevalent in Central and South American popula proach and explored whether study methodology moderates the tions, less prevalent in European Ancestry populations, and least magnitude of effect size. Li et al. [2006] explored the heterogeneity prevalent in Asian populations [Van Tol et al., 1991; Chang in effect sizes between association studies using study design et al., 1996]. (case–control vs. family-based) as a moderating variable and Despite an overall association between the DRD4 7R allele and found that case–control studies had a significantly higher mean AD/HD, a recent meta-analysis [Gizer et al., 2009] found that the effect size than family-based studies. The authors suggest that magnitude of the association varies substantially across studies. such differences may be the result of population stratification Though many different potential sources of heterogeneity exist, [Cardon and Bell, 2001] or differences between subjects researchers have focused primarily on AD/HD subtype and, to a ascertained from case–control and family-based study designs lesser extent, study methodology. [West et al., 2002].

Factors Accounting for Heterogeneity in the Summary and Hypotheses Association of DRD4 7R Allele With AD/HD Given the differences in samples, assessment methodology and AD/HD subtype. According to Sagvolden et al. [2005], hypo- study design, systematic variation in observed effect sizes [Gizer functioning in the mesolimbic dopaminergic pathway is related to et al., 2009] is likely to result from a combination of sample and hyperactivity–impulsivity, and hypofunctioning in the mesocort study characteristics. Building from previous meta-analyses that ical dopaminergic pathway is related to poor executive functioning have found significant variation in the association between AD/HD and attentional processes. Given that the DRD4 7R allele is impli and the DRD4 7R allele [Li et al., 2006; Gizer et al., 2009], the cated in the hypofunctioning of both dopaminergic pathways purpose of this study was to examine how certain sample and [Missale et al., 1998], it is likely that the DRD4 7R allele would study variables may moderate the association between DRD4 7R give rise to both inattention and hyperactivity–impulsivity symp allele and AD/HD. Of particular interest is the distribution of AD/ toms. Therefore, the DRD4 7R allele may be more strongly associ HD subtypes in identified samples. Thus, based on AD/HD etio ated with individuals with AD/HD-C compared to AD/HD-I. logical theory [Sagvolden et al., 2005], the following hypothesis was Furthermore, youth with AD/HD and the 7R allele tend to have made: a quicker response time [Manor et al., 2002; Langley et al., 2004] and a more consistent reaction time to target stimuli An increase in the proportion of individuals with AD/HD-C in the than AD/HD youth without the 7R allele [Swanson et al., AD/HD sample would be associated with an increase in magni 2000; Manor et al., 2002; Bellgrove et al., 2005]. Therefore, it is tude of association between AD/HD and the DRD4 7R allele. unlikely that the DRD4 7R allele is associated with AD/HD-I, as Conversely, an increase in the proportion of AD/HD-I in the AD/ a subset of individuals with AD/HD-I has a slowed reaction time HD sample will be associated with a decrease in magnitude of to target stimuli [e.g., Derefinko et al., 2008]. However, findings association between AD/HD and the DRD4 7R allele. from studies that have examined the differential association between the DRD4 7R allele and AD/HD subtypes and AD/HD In addition, several exploratory moderators of the relationship symptom dimensions have been inconsistent [Rowe et al., 1998, between DRD4 7R allele and AD/HD were examined. Given that 2001; McCracken et al., 2000; Todd et al., 2001; Frank et al., additional studies have been published since Li et al. [2006], the 2004]. Similarly, findings on the association between single nucle moderator of study design (case–control and family-based) was re otide polymorphisms in the DRD4 promoter region and AD/HD examined. The allele frequency in cases and controls, the mean age symptom dimensions have also been inconsistent [Lasky-Su et al., of the AD/HD sample, the proportion of males in the AD/HD 2007, 2008]. sample, diagnostic classification system, and sample ethnicity were A few researchers have examined the association of the DRD4 7R also examined as potential moderators to the relationship between allele and AD/HD with the presence or absence of oppositional DRD4 7R and AD/HD. MATERIALS AND METHODS analysis but included in a moderating analysis as the proportion of Literature Search AD/HD subtypes could not be calculated in Hawi et al. [2000]. One study was excluded because the control sample did not consist of First, computer searches were conducted using the PubMed and healthy individuals [Ballon et al., 2007]. Note that a number of PsycInfo search engines. Keywords associated with AD/HD phe included case–control studies reported family-based association notype (ADHD, inattention and hyperactivity) were crossed with ORs [Hawi et al., 2000; Holmes et al., 2000; Mill et al., 2001; Roman words associated with the DRD4 7R allele (DRD4, D4DR, dopa- et al., 2001; Gornick et al., 2007]. In these studies, only the data from mine receptor) to identify relevant studies that were published the case–control association analyses were reported as they were between January 1990 and July 2009. The author reviewed abstracts associated with more precise sample information, included and if the study included an AD/HD or related sample (e.g., more participants, and are comparable to family-based results hyperkinetic disorder) and had genotyped DRD4 then the full-text [Evangelou et al., 2006]. article were retrieved. Next, the reference section of each full text- A total of 27 studies and 28 samples were included in the overall article, including previous meta-analyses [Faraone et al., 2001, meta-analysis. Sixteen case–control and 12 family-based samples 2005; Maher et al., 2002; Li et al., 2006; Gizer et al., 2009], were were included (see Tables IA and IB). AD/HD was diagnosed reviewed to find additional studies that were not identified in the according to DSM-IV criteria in the majority of studies. Smith computer search. All together, this approach yielded 44 studies that et al. [2003] did not use a formal diagnostic classification system but provided data on the association between AD/HD and DRD4 7R their sample approximated DSM-IV criteria for AD/HD-C or allele. Using a developed protocol (available upon request), data ADHD-HI. Curran et al. [2001] used developmentally deviant from each study was extracted on two separate occasions by this scores on a brief behavioral questionnaire. DSM-III criteria were study’s author. Next, the separate protocols for each study were employed in three studies [Comings et al., 1999; Maher et al., 2002; compared, and the corresponding author(s) for each identified El-Faddagh et al., 2004]. Johansson et al. [2008] used the ICD-10 study were sent the protocol, asked to clarify any discrepancies, and and made DSM-IV modifications allowing the sample to meet for asked to provide supplementary information including raw data to AD/HD-I. Swanson et al. [1998] required participants to meet both compute the effect size and values for moderating variables of DSM-IV and ICD-10 criteria. interest. No additional studies were identified in correspondence Some studies applied additional inclusion criteria including: (1) with corresponding authors. meeting criteria for DSM-IV AD/HD-C [LaHoste et al., 1996; Tahir et al., 2000; Carrasco et al., 2006]; (2) a therapeutic response to Inclusion Criteria stimulant medication [Swanson et al., 1998; Sunohara et al., 2000 A study was included in the meta-analysis if it met all of the Irvine Sample]; (3) absence of comorbid psychiatric disorders following criteria: (1) presented an association analysis between excluding ODD [Swanson et al., 1998; Muglia et al., 2000; Sunohara the DRD4 7R allele and AD/HD using either case–control or family- et al., 2000 both Samples]; and (4) male gender [Swanson et al., based methods; (2) included (or the author provided) sufficient 1998]. Common exclusionary criteria included low IQ or the data to calculate an odds ratio (OR) and variance for the association presence of pervasive developmental disorder or neurological between AD/HD and the DRD4 7R allele; (3) reported data from an disorder. independent sample or was the largest dataset in a set of studies with The majority of AD/HD samples were clinic-referred; three overlapping samples; and (4) case–control studies employed samples were community-based [Curran et al., 2001; Todd et al., healthy individuals for their control sample. For studies that used 2001; El-Faddagh et al., 2004]. In addition, most samples were both case–control and family-based methods, results from child-based; two adult samples were included [Muglia et al., 2000; – case–control analyses were reported. Finally, for moderating anal Johansson et al., 2008]. In case control studies, control samples yses, the study needed to provide a value for the relevant moderating were selected using a variety of different methods including healthy variable. For non-independent samples, the largest sample that blood donors, healthy siblings, paternity testing services and provided a value for the moderating variable was included. matched controls from epidemiological studies. In several studies, AD/HD was not formally assessed in the control sample; thus, some Identified Studies control subjects may have met diagnostic criteria for the disorder. Of the 44 identified studies, a total of 17 studies were excluded from the primary analysis. Five studies conducted with Asian popula Effect Size and Within Study Variance tions were excluded due to the absence of 7R alleles [Qian et al., In case–control studies the OR is the ratio of the odds of having the 2004; Brookes et al., 2005; Kim et al., 2005; Leung et al., 2005; Cheuk DRD4 7R allele to non-7R alleles in the ADHD group compared to et al., 2006]. Four studies were excluded because the OR corre controls. Similarly, the OR for haplotype-based haplotype relative sponding to the association between the 7R allele and AD/HD could risk (HHRR) studies was the ratio of the odds of parents transmit not be computed [Manor et al., 2002; Bhaduri et al., 2006; Brookes ting the 7R allele to non-7R allele transmissions to AD/HD cases, et al., 2006; Monuteaux et al., 2008]. Seven studies were excluded compared to the non-transmission of the DRD4 7R and non-7R due to partially overlapping samples with larger samples [Smalley alleles. For transmission disequilibrium test (TDT) studies the et al., 1998; Faraone et al., 1999; McCracken et al., 2000; Holmes method reported by Lohmueller et al. [2003] was followed. Specifi et al., 2002; Kirley et al., 2002; Grady et al., 2003; Johnson et al., cally, to compute the OR for TDT studies, the frequency that a 2008]. Lowe et al. [2004] was also excluded from the primary heterozygous parent passed on the DRD4 7R allele to their affected TABLE IA. Effect Size and Sample Characteristics for Case–Control Association Studies of the DRD4 7R Allele and AD/HD

References Country ADHD N Control N Age Male CT IT DRD4 7R frequency (ADHD) DRD4 7R frequency (control) OR (95% CI) LaHoste et al. [1996] US 39 39 — — 1.00 0.00 0.28 0.12 3.01 (1.29–7.05) Rowe et al. [1998] US 107 58 9.4 0.70 0.65 0.35 0.24 0.13 2.16 (1.16–4.04) Comings et al. [1999] US 52 737 — — — — 0.18 0.14 1.41 (0.84–2.37) Hawi et al. [2000] Ireland 99 88 — 0.86 — — 0.24 0.26 0.93 (0.58–1.49) Holmes et al. [2000] UK 129 442 — — 0.76 0.08 0.22 0.13 1.89 (1.33–2.70) Muglia et al. [2000] Canada 66 66 34.3 0.56 — — 0.21 0.10 2.46 (1.21–5.00) Curran et al. [2001] UK 133 91 — — — — 0.26 0.14 2.14 (1.30–3.52) Mill et al. [2001] UK 132 189 10.4 — 0.91 0.02 0.21 0.14 1.69 (1.11–2.56) Roman et al. [2001] Brazil 66 100 10.1 0.86 0.77 0.16 0.25 0.17 1.63 (.95–2.79) Smith et al. [2003] US 105 68 — — — — 0.20 0.20 0.98 (0.57–1.68) El-Faddagh et al. [2004] Germany 24 231 4.5 0.71 — — 0.33 0.19 2.19 (1.15–4.16) Frank et al. [2004] US 81 24 — 0.80 — 0.38 0.30 0.25 1.26 (0.61–2.63) Carrasco et al. [2006] Chile 26 25 — 0.92 1.00 0.00 0.19 0.14 1.46 (0.51–4.20) Gornick et al. [2007] US 166 282 9.0 0.53 0.94 0.06 0.23 0.17 1.45 (1.03–2.03) Johansson et al. [2008] Norway 358 340 33.9 0.51 0.21 0.17 0.22 0.24 0.90 (0.70–1.15) Martinez-Levy et al. [2009] Mexico 105 84 14.3 — — — 0.30 0.35 0.77 (0.50–1.20)

Age ¼ mean age of the AD/HD sample; Male ¼ proportion of males in the AD/HD sample; CT ¼ proportion of individuals with ADHD-C in the AD/HD sample; IT ¼ proportion of individuals with AD/HD-I in the AD/HD sample.

TABLE IB. Effect Size and Sample Characteristics for Family-Based Association Studies of the DRD4 7R Allele and AD/HD

References Method Country ADHD N Triads Dyads Age Male CT IT OR (95% CI) Swanson et al. [1998] HHRR US 52 52 0 — 1.00 — — 2.08 (1.06–4.06) Kotler et al. [2000] HHRR Isreal 47 — — 9.89 — 0.87 0.13 0.47 (0.22–.99) Lunetta et al. [2000] TDT US 17 4 13 — — — — 1.50 (0.58–3.87) Sunohara et al. [2000], Irvine TDT US 59 21 38 — — — — 1.00 (0.57–1.74) Sunohara et al. [2000], Toronto TDT Canada 88 75 13 — — — — 1.52 (0.03–2.47) Tahir et al. [2000] TDT Turkey 29 26 3 — — 1.00 0.00 1.90 (0.88–4.09) Todd et al. [2001] TDT US 201 201 0 — 0.56 0.36 0.58 1.11 (0.77–1.59) Maher et al. [2002] TDT US 33 33 0 — — — — 1.13 (0.43–2.92) Arcos-Burgos et al. [2004] TDT Colombia — — — — — — — 1.13 (0.43–2.92) Kustanovich et al. [2004] TDT US 535 535 0 11.0 .74 0.50 0.43 1.21 (0.94–1.56) Lowe et al. [2004] TDT Ireland 178 — — — — 0.63 0.12 1.35 (0.88–2.07) Bakker et al. [2005] TDT Netherlands 236 231 5 — — — — 1.06 (0.76–1.48) Niederhofer et al. [2008] TDT Germany 49 36 13 8.4 0.88 0.59 0.33 0.92 (0.40–2.08)

Age ¼ mean age of the AD/HD sample; Male ¼ proportion of males in the AD/HD sample; CT ¼ proportion of individuals with AD/HD-C in the sample; IT ¼ proportion of individuals with AD/HD-I in the sample. offspring was divided by the frequency that the allele was not passed to their affected offspring. Such a method is considered asymptoti cally equivalent to calculating an OR in a case–control study, as in a large population of control participants the expected transmission ratio would approach 50:50 [Lohmueller et al., 2003]. The ORs for both study types were then converted to log ORs and the log OR variance within case–control and HHRR studies was calculated by summing the reciprocals of the 7R allele and non-7R allele frequen cies in cases and controls and for transmitted and non-transmitted alleles, respectively. The variance in TDT studies was calculated assuming a large control sample. Thus, the sum of the reciprocals for the frequency of the transmitted risk allele and non-transmis sion of the risk allele were considered asymptotically equivalent to the sum of the reciprocal from each cell of a case–control study [Lohmueller et al., 2003].

Statistical Analysis Meta-analysis of case–control, HHRR and TDT studies was con ducted using Comprehensive Meta-Analysis (Version 2.0, BIO STAT, Englewood, NJ). Given that the samples and methods employed between studies are variable, it is assumed that the effect size from each study is a sample from a distribution of true effects FIG. 1. Forest plot of OR and pooled OR for association studies [Borenstein et al., 2009]. Thus, a random effects model was used to between AD/HD and the DRD4 7R allele. estimate the mean of all relevant true effects [Borenstein et al., 2009] and the mean effect of moderator variables. Compared to a fixed effects model, a random effects model allows inferences about model, g0 represents the estimated effect size when W1j is equal to population parameters including the effect size and regression zero, and g1 indicates the amount of change in the effect size for a coefficients. The presence of heterogeneity in effect sizes between one-unit increase in W1j [Feinn et al., 2005]. Two-tailed z-tests were studies was tested using the c2-based Q-statistic. Heterogeneity was used to assess the impact of the moderating variables on effect size. also measured with T2 and t which are estimates of the variance and An analogous index for proportion of variance explained was 2 ¼ 2 = 2 standard deviation of true effects using the DerSimonian and Laird calculated; (Ranalog Texplained TTotal ) to describe the proportion [1986] method. The amount of heterogeneity was quantified using of systematic, between studies variance that is explained by the I2, which measures the proportion of total variation that reflects real presence of the moderating variable [Borenstein et al., 2009]. Alpha differences in the variability between studies in observed effect size was set to 0.05 for hypothesized analyses and reduced to 0.007 for [Borenstein et al., 2009]. exploratory meta-regression and sub-group analyses using a Bon Subgroup analysis and meta-regression analyses were utilized to feronni adjustment. examine the influence of categorical and continuous moderating variables on observed effect size. Subgroup analyses utilized a RESULTS mixed-effects model, where T2 was calculated separately using a random effects model within subgroups and using a fixed effects See Tables IA and IB for a summary of included studies. Twenty- model between subgroups [Borenstein et al., 2009]. If a subgroup eight independent samples met inclusion criteria to estimate the had fewer than five studies, then T2 was pooled across subgroups mean effect size for the association between the DRD4 7R allele and using a random-effects model, consistent with the recommenda AD/HD. A total of 1,688 identified AD/HD cases and 2,864 control tion from Borenstein et al. [2009]. A Q-test was used to test for subjects drawn from case–control studies were included in the heterogeneity across subgroups. meta-analysis. In family-based studies, approximately 1,3461 af A method of moments meta-regression was used to calculate the fected individuals participated. Fewer individuals in each sample following model. Following notation by Raudenbush and Bryk were included in TDT analyses as they require a parent to be ¼ þ þ þ [2002]:dj g0 g1W1j uj ejwhere dj is the odds ratio of heterozygous for the allele of interest. study j, g0 and g1 are regression coefficients; W1j is a study In the absence of a moderating variable, the unconditional model characteristic predicting effect sizes (e.g., proportion of AD/HD produced a log OR of 0.29 for g0. This corresponds to an OR of C individuals in a sample); and uj is the between studies random 1.33 (95% CI ¼ 1.16–1.53, z ¼ 4.04, P ¼ 0.00005; Fig. 1). Neither error which is not predicted by the study characteristic for which we visual inspection of a funnel plot nor Egger’s funnel plot statistic 2 2 assume uj � N (0, t ) and t is the between studies variance and ej is 1 sampling error associated with the estimate (i.e., dj) of the popula The number 1,346 reflects the combined sample size from family-based tion effect size d, for study j, for which we assume ej � N (0, Vj). Here studies with the exception of Arcos-Burgos et al. 2004 as the number of Vj is considered ‘‘known’’ and is the sampling variance for dj. In this individuals with AD/HD from this study was not reported. TABLE II. Study Characteristics Examined as Potential Moderating Variables of Effect Size

Moderator g1 SE z dfresidual P Transformeda proportion of AD/HD-C in AD/HD sample 1.47 0.55 2.67 13 0.008 Proportion of AD/HD-C in AD/HD sample 0.85 0.28 3.02 13 0.003 Transformedb proportion of AD/HD-I in AD/HD sample -0.88 0.38 -2.28 13 0.023 Proportion of AD/HD-I in AD/HD sample -0.90 0.43 -2.09 13 0.037 DRD4 7R allele frequency in AD/HD sample 0.99 2.72 0.37 14 0.715 7R Allele frequency in control sample -5.02 0.92 -5.48 14 0.000 Proportion of AD/HD males in AD/HD sample 0.52 0.81 0.64 8 0.523 Log mean AD/HD age -0.18 0.21 -0.85 10 0.396

The intercept for each analysis is not presented as the unconditional model demonstrated an overall association between the risk allele and the disorder, and the intercept for all moderating analyses varied substantially. In addition, given that the coverage of some covariates near the intercept were lacking, estimates of the intercept were considered extrapolations beyond available data.a1/(2 - X). bHX.

[P ¼ 0.062; Egger et al., 1997] indicated the presence of publication bias. In addition, Rosenthal’s fail-safe N suggests that 227 additional studies with a mean OR of 1.0 would need to be added to the analysis before the effect would be no longer significant. ORs ranged from 1.30 to 1.36 and remained statistically significant when each study was removed from the analysis, one at time, indicating that no one study produced the significant overall effect. In the unconditional model, the variability in observed effect size between studies was greater than would be expected if each study shared a common effect (Q ¼ 54.24; P ¼ 0.001). Furthermore, I2 ¼ 50.22 suggests that approximately half of the observed variability in effect size was systematic in nature. This indicates that moderating variables, such as AD/HD subtype, may account for the heteroge FIG. 2. Random-effects meta-regression analysis of study log OR neity in effect sizes between studies. In order to test for this, the on the transformed proportion AD/HD-C individuals in the AD/HD unconditional model was expanded to include moderating varia sample. The size of each data point reflects the relative weight of each study. bles. Table II displays a summary of the results from these analyses. To test the primary hypothesis that the DRD4 7R allele is more strongly associated with AD/HD-C compared to other AD/HD AD/HD-I individuals in the AD/HD sample was associated with a subtypes, the proportion of AD/HD-C individuals within the AD/ decreased log OR (see Fig. 3). This finding was consistent with the HD sample was entered into the equation as a moderating variable. effect of the non-transformed proportion of the AD/HD-I variable Given the proportional nature of this variable and that multiple on observed log OR. studies included only AD/HD-C individuals, the proportion of AD/ Next, to explore whether study design accounted for variability HD-C individuals within reporting studies was negatively skewed, in effect size, studies were dichotomized into either case–control or thus the reciprocal of 2 - X was applied to normalize the distribu family-based design. Though effect sizes between studies were not tion.2 Increases in the transformed proportion of AD/HD-C in significantly different (Q ¼ 2.76; P ¼ 0.097), case–control studies dividuals in the AD/HD sample predicted increases in the observed had a larger mean effect size and a larger proportion of variability log OR (see Fig. 2). The transformed proportion of AD/HD-C in observed effects (OR ¼ 1.46; 95% CI ¼ 1.19–1.79; z ¼ 3.66; individuals in the AD/HD sample accounted for 95% of the P ¼ 0.0002; I2 ¼ 62.04) compared to family-based analyses systematic variability in the observed ORs between studies. This (OR ¼ 1.17; 95% CI ¼ 1.00–1.38; z ¼ 1.94; P ¼ 0.052; I2 ¼ 13.04). was consistent with the untransformed proportion of AD/HD-C. Exploratory meta-regression analyses examined variability in These findings were mirrored by the proportion of AD/HD-I case–control studies. Allele frequency in the AD/HD sample did individuals in the AD/HD sample. The distribution of the propor not predict observed log OR, whereas increases in the allele fre tion of the AD/HD-I variable was positively skewed and a square quency in the control sample was associated with decreases in the root transformation was applied to normalize the distribution.3 observed log OR (see Fig. 4). Within case–control studies, the allele Increases in the square root transformation of the proportion of frequency in the control sample accounted for approximately 100%4 of the systematic variability in observed effect size between 2Where X is the proportion of individuals diagnosed with AD/HD-C. ¼ ¼ - 4 2 When X 1 the proportion was revised to X1 2v 1/2v; where v is the The Ranalog theoretically ranges from 0 to 1 in the population; AD/HD sample size. however, sampling error may cause the index to fall outside this 3 2 ¼ : When the proportion of individuals with AD/HD-I was zero, the range. In the present study, Ranalog 1 30, the value was set to 1.0 proportion was revised to 1/2v; where v is the AD/HD sample size. [Borenstein et al., 2009]. allele increases the odds that an individual is diagnosed with AD/ HD by 33%. This is in line with recent meta-analyses on the same topic that used slightly different inclusion criteria [Faraone et al., 2005; Li et al., 2006; Gizer et al., 2009]. Consistent with recent meta-analyses [Li et al., 2006; Gizer et al., 2009], the magnitude of association varied significantly between studies. In an extension of previous topical meta-analyses and consistent with this study’s primary hypothesis, increases in the proportion of AD/HD-C individuals within the AD/HD sample were associated with an increase in the magnitude of association between the DRD4 7R allele and AD/HD. Findings suggest that including only AD/HD-C individuals in an analysis would increase FIG. 3. Random-effects meta-regression analysis of study log OR the OR from 1.22 to 1.79, assuming 55% of the AD/HD sample met on the transformed proportion of AD/HD-I individuals in the AD/ criteria for AD/HD-C [Lahey et al., 1994]. Conversely, as the HD sample. The size of each data point reflects the relative proportion of AD/HD-I individuals in the AD/HD sample in weight of each study. creased, the observed OR decreased. Taken together, results were consistent with the hypothesis that the DRD4 7R allele is more strongly associated with AD/HD-C compared to AD/HD-I. The relative proportion of the two predominant AD/HD subtypes in the AD/HD sample accounted for the majority of systematic variability between reporting studies. Together, findings suggest that hypofunctioning in mesocortical and mesolimbic dopaminergic path ways may better characterize the etiology of AD/HD-C compared to AD/HD-I [Sagvolden et al., 2005]. This is in line with findings that the DAT1 10-repeat allele is more strongly associated with AD/HD C than AD/HD-I [e.g., Waldman et al., 1998]. Furthermore, this meta-analysis suggests that reducing the AD/HD phenotypic het erogeneity may lead to the discovery of AD/HD etiological subtypes or possibly distinct disorders [e.g., Milich et al., 2001]. No other sample characteristics accounted for a significant proportion of FIG. 4. Random-effects meta-regression analysis of case–control true between studies variance. study log OR on the proportion of the DRD4 7R allele frequency in In addition to sample characteristics, the association between the control sample. The size of each data point reflects the study methodology and magnitude of effect size was explored. In relative weight of each study. contrast to Li et al. [2006], case–control studies did not have a significantly larger effect size than family-based studies, but there studies, whereas separately, the allele frequency in the AD/HD was a trend in this direction. Given that results of case–control population only accounted for 19%. studies may be biased by population stratification [Cardon and Bell, Exploratory analyses were also conducted to examine if other 2001], the relationship between 7R allele frequencies in the AD/HD variables were associated with the variability in effect sizes between and control samples and OR was examined. The allele frequency in studies. In subgroup analyses, studies drawing from Mexico, Brazil, the AD/HD sample was not related to the observed OR whereas the Chile and Colombia were not significantly different from other allele frequency in the control population did predict observed OR studies (Q ¼ 0.07; P ¼ 0.407). In addition, studies using DSM-IV and accounted for close to all of the systematic variability of were not significantly different than studies using other diagnostic observed ORs between case–control studies. Given that the DRD4 approaches (Q ¼ 0.88; P ¼ 0.348). In meta-regression, neither the allele frequencies vary substantially across ethnic groups [e.g., proportion of males nor the log of the mean age of the AD/HD Chang et al., 1996] and that the association between the 7R allele sample5 predicted variability in observed effect size between studies. and AD/HD may be dynamic [Shaw et al., 2007], the appropriate ness of the control sample in case–control studies is of critical DISCUSSION importance. Failure to appropriately control for such character istics may have influenced the variability in findings. The present meta-analysis examined the association of the DRD4 These findings provide guidance for further study. First, indi 7R allele with AD/HD by combining 16 case–control and 12 family- cators for the quality of AD/HD phenotyping are underreported in based samples, which together included over 3,000 AD/HD cases the literature. Future studies would be enhanced by: including and over 2,800 controls. The overall mean OR between the DRD4 information related to the psychometric properties of their assess 7R allele and AD/HD was 1.33. This indicates that the DRD4 7R ments; describing how assessments are combined to form a diag nosis; reporting how many cases and controls did not meet criteria 5Mean age of the AD/HD sample was severely positively skewed. Thus a for study inclusion; and by presenting descriptive sample statistics log transformation was applied to normalize the distribution. for both cases and controls. In terms of genotyping, few studies adequately reported tests of Hardy–Weinberg equilibrium [Gizer future research should look to elucidate the neurobiological under et al., 2009]. Future studies should report such information as it pinnings of severe inattentive symptomatology in the absence of may help to explain variability between study findings. Though this hyperactivity–impulsivity so that the genetic underpinnings of this meta-analysis included a relatively large number of samples, values symptom presentation may be better understood. To the extent that for covariates of interest were often unavailable. For instance, like AD/HD-C and AD/HD-I (with few or no hyperactivity–impulsivity AD/HD subtype, differences in the rates of Conduct Disorder symptoms) have unique etiological pathways, there would be within and between samples may account for heterogeneity in sufficient evidence to categorize these AD/HD subtypes as separate candidate gene association studies [Thapar et al., 2006]; however, disorders [e.g., Milich et al., 2001]. too few studies presented the rate of conduct disorder within their sample to be examined in this meta-analysis. In addition, values of ACKNOWLEDGMENTS covariates for TDT studies were attained from summary statistics that were based on the sample at-large and not only individuals I would like to thank Dr. Arthur Anastopoulos, Dr. Thomas Kwapil, included in the TDT analysis. Future research should stratify results Jessica Benson, Allison Coville, Jessica Kaczorowski, Sarah O based on potential moderating variables [Li et al., 2006] from the ’Rourke, Nicole Schatz, and Jennifer Sommer for their helpful sample or sub-sample for which the results are based. comments on this article. I would also like to thank These findings also need to be considered in light of the limi Dr. Mauricio Acros-Burgos, Dr. Allison Ashley-Koch, Dr. Cathy tations inherent in meta-regression analysis and subgroup analysis Barr, Dr. Sarah Curran, Dr. Ian Gizer, Dr. Stephen Faraone, [see Thompson and Higgins, 2002 for a review]. First, values for Dr. Carlos Cruz-Fuentes, Dr. Kathryn Lunetta, Gabriela moderating variables were not randomly assigned to studies; thus, Martinez-Levy, Dr. Michael Monuteaux, Dr. Eda Tahir, relationships may be related to confounding bias. Second, meta Dr. Anita Thapar, Dr. Francisco Rothhammer, and Dr. Irwin regression analyses examining sample characteristics deal with Waldman for their correspondence related to this article. sample averages between studies and not individual level data within studies, thus observed relationships between studies are not REFERENCES necessarily found within studies (e.g., ecological bias). For instance, though some evidence suggests that the relationship between AD/ Arcos-Burgos M, Castellanos FX, Konecki D, Lopera F, Pineda D, Palacio HD and the DRD4 7R allele may reduce with age [Shaw et al., 2007], JD, Rapoport JL, Berg K, Bailey-Wilson J, Muenke M. 2004. Pedigree limited between study variability may have masked this association, disequilibrium test (PDT) replicates association and linkage between which may be present within studies. This limitation can be DRD4 and ADHD in multigenerational and extended pedigrees from a – attenuated through large collaborative efforts or by data stratified genetic isolate. Mol Psychiatry 9:252 259. by moderating variables of interest. Finally, a practical limitation of Bakker S, van der Meulen E, Oteman N, Schelleman H, Pearson P, Buitelaar meta-analysis and especially meta-regression is that access to J, Sinke R. 2005. DAT1, DRD4, and DRD5 polymorphisms are not observed effect size, its variance, and values for covariates are associated with ADHD in Dutch families. Am J Med Genet Part B 132B:50–52. necessary. Despite these limitations, such findings support the argument Ballon N, Leroy S, Roy C, Bourdel MC, Olie JP, Charles-Nicolas A, Krebs MO, Poirier MF. 2007. Polymorphisms TaqI A of the DRD2, BalI of the [e.g., Milich et al., 2001] that AD/HD-C and AD/HD-I (with few to DRD3, exon III repeat of the DRD4, and 3’ UTR VNTR of the DAT: no hyperactive–impulsive symptoms) may result from distinct Association with childhood ADHD in male African-Caribbean cocaine etiological pathways. Further delineating samples based on symp dependents? Am J Med Genet Part B 144B:1034–1041. toms related to Sluggish Cognitive Tempo [e.g., McBurnett et al., Bellgrove MA, Hawi Z, Lowe N, Kirley A, Robertson IH, Gill M. 2005. 2001] may allow for greater specificity in identifying shared and DRD4 gene variants and sustained attention in attention deficit hyper distinct pathways to distinct behavioral phenotypes. These findings activity disorder (ADHD): Effects of associated alleles at the VNTR and – also suggest that reducing the phenotypic variability within AD/HD 521 SNP. Am J Med Genet Part B 136B:81 86. may help to increase statistical power to detect candidate genes in Bhaduri N, Das M, Sinha S, Chattopadhyay A, Gangopadhyay PK, Chaud molecular genetic studies. huri K, Singh M, Mukhopadhyay K. 2006. Association of dopamine D4 By pooling summary data from a relatively large number of receptor (DRD4) polymorphisms with attention deficit hyperactivity disorder in Indian population. Am J Med Genet Part B 141B:61–66. association studies, evidence suggests that the DRD4 7R allele is more strongly related to AD/HD-C compared to AD/HD-I. These Borenstein M, Hedges LV, Higgins JP, Rothstein HR. 2009. Introduction to

findings suggest that etiological theories of AD/HD, which are meta-analysis. Chichester, West Sussex, UK/Hoboken: John Wiley & Sons. 421. pp. primarily based on AD/HD-C individuals, may fall short in identi fying appropriate candidate susceptibility genes for AD/HD-I. Brookes KJ, Xu X, Chen CK, Huang YS, Wu YY, Asherson P. 2005. No evidence for the association of DRD4 with ADHD in a Taiwanese Additionally, given that the majority of genetic studies of AD/HD population within-family study. BMC Med Genet 6:31–36. are conducted on clinical samples, which are disproportionally diagnosed with AD/HD-C [Lahey et al., 1994], such investigations Brookes KJ, Mill J, Guindalini C, Curran S, Xu X, Knight J, Chen CK, Huang YS, Sethna V, Taylor E, Chen W, Breen G, Asherson P. 2006. A common may lack sufficient power to identify unique vulnerability genes haplotype of the dopamine transporter gene associated with attention associated with AD/HD-I. Therefore, exploratory genome wide deficit/hyperactivity disorder and interacting with maternal use of alco association studies may have the potential to identify genes that are hol during pregnancy. Arch Gen Psychiatry 63:74–81. more replicable and have a stronger association [Gizer et al., 2009] Cardon L, Bell J. 2001. Association study designs for complex diseases. for this group of individuals. To guide candidate gene studies, Nature Rev Genet 2:91–99. Carrasco X, Rothhammer P, Moraga M, Henriquez H, Chakraborty R, Gaub M, Carlson CL. 1997. Behavioral characteristics of DSM-IV ADHD Aboitiz F, Rothhammer F. 2006. Genotypic interaction between DRD4 subtypes in a school-based population. J Abnorm Child Psychol 25: and DAT1 loci is a high risk factor for attention-deficit/hyperactivity 103–111. disorder in Chilean families. Am J Med Genet Part B 141B:51–54. Gelernter J, Kennedy JL, van Tol HH, Civelli O, Kidd KK. 1992. The D4 Chang FM, Kidd JR, Livak KJ, Pakstis AJ, Kidd KK. 1996. The world-wide dopamine receptor (DRD4) maps to distal 11p close to HRAS. Genomics distribution of allele frequencies at the human dopamine D4 receptor 13:208–210. locus. Hum Genet 98:91–101. Gizer IR, Ficks C, Waldman ID. 2009. Candidate gene studies of ADHD: A Cheuk DK, Li SY, Wong V. 2006. Exon 3 polymorphisms of dopamine D4 meta-analytic review. Hum Genet 126:51–90. receptor (DRD4) gene and attention deficit hyperactivity disorder in Gornick MC, Addington A, Shaw P, Bobb AJ, Sharp W, Greenstein D, Chinese children. Am J Med Genet Part B 141B:907–911. Arepalli S, Castellanos FX, Rapoport JL. 2007. Association of the dopa- Comings DE, Gonzalez N, Wu S, Gade R, Muhleman D, Saucier G, Johnson mine receptor D4 (DRD4) gene 7-repeat allele with children with P, Verde R, Rosenthal RJ, Lesieur HR, Rugle LJ, Miller WB, MacMurray attention-deficit/hyperactivity disorder (ADHD): An update. Am J Med JP. 1999. Studies of the 48 bp repeat polymorphism of the DRD4 gene in Genet Part B 144B:379–382. impulsive, compulsive, addictive behaviors: Tourette syndrome, ADHD, Grady DL, Chi HC, Ding YC, Smith M, Wang E, Schuck S, Flodman P, pathological gambling, and substance abuse. Am J Med Genet Part B Spence MA, Swanson JM, Moyzis RK. 2003. High prevalence of rare 88B:358–368. dopamine receptor D4 alleles in children diagnosed with attention-deficit Curran S, Mill J, Sham P, Rijsdijk F, Marusic K, Taylor E, Asherson P. 2001. hyperactivity disorder. Mol Psychiatry 8:536–545. QTL association analysis of the DRD4 exon 3 VNTR polymorphism in a Hawi Z, McCarron M, Kirley A, Daly G, Fitzgerald M, Gill M. 2000. No population sample of children screened with a parent rating scale for association of the dopamine DRD4 receptor (DRD4) gene polymor ADHD symptoms. Am J Med Genet Part B 105B:387–393. phism with attention deficit hyperactivity disorder (ADHD) in the Irish Derefinko KJ, Adams ZW, Milich R, Fillmore MT, Lorch EP, Lynam DR. population. Am J Med Genet Part B 96B:268–272. 2008. Response style differences in the inattentive and combined subtypes Holmes J, Payton A, Barrett JH, Hever T, Fitzpatrick H, Trumper AL, of attention-deficit/hyperactivity disorder. J Abnorm Child Psychol Harrington R, McGuffin P, Owen M, Ollier W, Worthington J, Thapar A. 36:745–758. 2000. A family-based and case-control association study of the dopamine DerSimonian R, Laird N. 1986. Meta-analysis in clinical trials. Control Clin D4 receptor gene and dopamine transporter gene in attention deficit Trials 7:177–188. hyperactivity disorder. Mol Psychiatry 5:523–530. DuPaul G, Anastopoulos A, Power T, Reid R, Ikeda M, McGoey K. 1998. Holmes J, Payton A, Barrett J, Harrington R, McGuffin P, Owen M, Ollier Parent ratings of attention-deficit/hyperactivity disorder symptoms: W, Worthington J, Gill M, Kirley A, Hawi Z, Fitzgerald M, Asherson P, Factor structure and normative data. J Psychopathol Behav Assess Curran S, Mill J, Gould A, Taylor E, Kent L, Craddock N, Thapar A. 2002. 20:83–102. Association of DRD4 in children with ADHD and comorbid conduct problems. Am J Med Genet Part B 114B:150–153. Egger M, Davey Smith G, Schneider M, Minder C. 1997. Bias in meta analysis detected by a simple, graphical test. BMJ 315:629–634. Johansson S, Halleland H, Halmoy A, Jacobsen KK, Landaas ET, Dram sdahl M, Fasmer OB, Bergsholm P, Lundervold AJ, Gillberg C, Hugdahl El-Faddagh M, Laucht M, Maras A, Vohringer L, Schmidt MH. 2004. K, Knappskog PM, Haavik J. 2008. Genetic analyses of dopamine related Association of dopamine D4 receptor (DRD4) gene with attention genes in adult ADHD patients suggest an association with the DRD5 deficit/hyperactivity disorder (ADHD) in a high-risk community sam microsatellite repeat, but not with DRD4 or SLC6A3 VNTRs. Am J Med ple: A longitudinal study from birth to 11 years of age. J Neural Transm Genet Part B 147B:1470–1475. 111:883–889. Johnson KA, Kelly SP, Robertson IH, Barry E, Mulligan A, Daly M, Lambert Evangelou E, Trikalinos TA, Salanti G, Ioannidis JP. 2006. Family-based D, McDonnell C, Connor TJ, Hawi Z, Gill M, Bellgrove MA. 2008. versus unrelated case-control designs for genetic associations. PLoS Absence of the 7-repeat variant of the DRD4 VNTR is associated with Genet 2:e123. drifting sustained attention in children with ADHD but not in controls. Faraone SV. 2008. Statistical and molecular genetic approaches to devel Am J Med Genet Part B 147B:927–937. opmental psychopathology. In: Hudziak JJ, editor. Developmental psy Kim YS, Leventhal BL, Kim SJ, Kim BN, Cheon KA, Yoo HJ, Badner J, chopathology and wellness: Genetic and environmental influences. Cook EH. 2005. Family-based association study of DAT1 and DRD4 Washington: American Psychiartic Publishing, Inc. pp. 245–265. polymorphism in Korean children with ADHD. Neurosci Lett 390: Faraone SV, Biederman J, Weiffenbach B, Keith T, Chu MP, Weaver A, 176–181. Spencer TJ, Wilens TE, Frazier J, Cleves M, Sakai J. 1999. Dopamine D4 Kirley A, Hawi Z, Daly G, McCarron M, Mullins C, Millar N, Waldman I, gene 7-repeat allele and attention deficit hyperactivity disorder. Am J Fitzgerald M, Gill M. 2002. Dopaminergic system genes in ADHD: Psychiatry 156:768–770. Toward a biological hypothesis. Neuropsychopharmacology 27: Faraone SV, Doyle AE, Mick E, Biederman J. 2001. Meta-analysis of the 607–619. association between the 7-repeat allele of the dopamine D(4) receptor Kirley A, Lowe N, Mullins C, McCarron M, Daly G, Waldman I, Fitzgerald gene and attention deficit hyperactivity disorder. Am J Psychiatry M, Gill M, Hawi Z. 2004. Phenotype studies of the DRD4 gene poly 158:1052–1057. morphisms in ADHD: Association with oppositional defiant disorder Faraone S, Perlis R, Doyle A, Smoller J, Goralnick J, Holmgren M, Sklar P. and positive family history. Am J Med Genet Part B 131B:38–42. 2005. Molecular genetics of attention-deficit/hyperactivity disorder. Biol Kotler M, Manor I, Sever Y, Eisenberg J, Cohen H, Ebstein RP, Tyano S. Psychiatry 57:1313–1323. 2000. Failure to replicate an excess of the long dopamine D4 exon III Feinn R, Nellissery M, Kranzler HR. 2005. Meta-analysis of the association repeat polymorphism in ADHD in a family-based study. Am J Med Genet of a functional serotonin transporter promoter polymorphism with Part B 96B:278–281. alcohol dependence. Am J Med Genet Part B 133B:79–84. Kustanovich V, Ishii J, Crawford L, Yang M, McGough JJ, McCracken Frank Y, Pergolizzi R, Perilla M. 2004. Dopamine D4 receptor gene and JT, Smalley SL, Nelson SF. 2004. Transmission disequilibrium testing attention deficit hyperactivity disorder. Pediatr Neurol 31:345–348. of dopamine-related candidate gene polymorphisms in ADHD: Confirmation of association of ADHD with DRD4 and DRD5. Mol interaction analysis for DRD4 and DAT1 genes in a group of Mexican Psychiatry 9:711–717. ADHD patients. Neurosci Lett 451:257–260. Lahey BB, Applegate B, McBurnett K, Biederman J, Greenhill L, Hynd GW, McBurnett K, Pfiffner L, Frick P. 2001. Symptom properties as a function of Barkley RA, Newcorn J, Jensen P, Richters J. 1994. DSM-IV field trials for ADHD type: An argument for continued study of sluggish cognitive attention deficit hyperactivity disorder in children and adolescents. Am J tempo. J Abnorm Child Psychol 29:207–213. Psychiatry 151:1673–1685. McCracken JT, Smalley SL, McGough JJ, Crawford L, Del’Homme M, Lahey BB, Pelham WE, Loney J, Lee SS, Willcutt E. 2005. Instability of the Cantor RM, Liu A, Nelson SF. 2000. Evidence for linkage of a tandem DSM-IV Subtypes of ADHD from preschool through elementary school. duplication polymorphism upstream of the dopamine D4 receptor gene Arch Gen Psychiatry 62:896–902. (DRD4) with attention deficit hyperactivity disorder (ADHD). Mol Psychiatry 5:531–536. LaHoste GJ, Swanson JM, Wigal SB, Glabe C, Wigal T, King N, Kennedy JL. 1996. Dopamine D4 receptor gene polymorphism is associated with Milich R, Balentine A, Lynam D. 2001. ADHD combined type and ADHD attention deficit hyperactivity disorder. Mol Psychiatry 1:121–124. predominantly inattentive type are distinct and unrelated disorders. Clin Psychol: Sci Pract 8:463–488. Langley K, Marshall L, van den Bree M, Thomas H, Owen M, O’Donovan M, Thapar A. 2004. Association of the dopamine D4 receptor gene 7 Mill J, Curran S, Kent L, Richards S, Gould A, Virdee V, Huckett L, Sharp J, repeat allele with neuropsychological test performance of children with Batten C, Fernando S, Simanoff E, Thompson M, Zhao J, Sham P, Taylor ADHD. Am J Psychiatry 161:133–138. E, Asherson P. 2001. Attention deficit hyperactivity disorder (ADHD) and the dopamine D4 receptor gene: Evidence of association but no Lasky-Su J, Banaschewski T, Buitelaar J, Franke B, Brookes K, Sonuga- linkage in a UK sample. Mol Psychiatry 6:440–444. Barke E, Ebstein R, Eisenberg J, Gill M, Manor I, Miranda A, Mulas F, Oades RD, Roeyers H, Rothenberger A, Sergeant J, Steinhausen HC, Missale C, Nash SR, Robinson SW, Jaber M, Caron MG. 1998. Dopamine Taylor E, Zhou K, Thompson M, Asherson P, Faraone SV. 2007. Partial receptors: From structure to function. Physiol Rev 78:189–225. replication of a DRD4 association in ADHD individuals using a statisti Monuteaux MC, Seidman LJ, Faraone SV, Makris N, Spencer T, Valera E, cally derived quantitative trait for ADHD in a family-based association Brown A, Bush G, Doyle AE, Hughes S, Helliesen M, Mick E, Biederman J. test. Biol Psychiatry 62:985–990. 2008. A preliminary study of dopamine D4 receptor genotype and Lasky-Su J, Lange C, Biederman J, Tsuang M, Doyle AE, Smoller JW, Laird structural brain alterations in adults with ADHD. Am J Med Genet Part N, Faraone S. 2008. Family-based association analysis of a statistically B 147B:1436–1441. derived quantitative traits for ADHD reveal an association in DRD4 with Muglia P, Jain U, Macciardi F, Kennedy J. 2000. Adult attention deficit inattentive symptoms in ADHD individuals. Am J Med Genet Part B hyperactivity disorder and the dopamine D4 receptor gene. Am J Med 147B:100–106. Genet Part B 96:273–277. Leung PW, Lee CC, Hung SF, Ho TP, Tang CP, Kwong SL, Leung SY, Yuen Niederhofer H, Menzel F, Gobel K, Hackenberg B, Richter R, Walter MH, ST, Lieh-Mak F, Oosterlaan J, Grady D, Harxhi A, Ding YC, Chi HC, Gross C, Huber M, Pycha R, Menzel HJ. 2008. A preliminary report of the Flodman P, Schuck S, Spence MA, Moyzis R, Swanson J. 2005. Dopamine dopamine receptor D(4) and the dopamine transporter 1 gene polymor receptor D4 (DRD4) gene in Han Chinese children with attention phism and its association with attention deficit hyperactivity disorder. deficit/hyperactivity disorder (ADHD): Increased prevalence of the 2 Neuropsychiatr Dis Treat 4:701–705. repeat allele. Am J Med Genet Part B 133B:54–56. O’Malley K, Harmon S, Tang L, Todd R. 1992. The rat dopamine D4 Levy F. 1991. The dopamine theory of attention deficit hyperactivity receptor: Sequence, gene structure, and demonstration of expression in disorder (ADHD). Aust NZ J Psychiatry 25:277–283. the cardiovascular system. New Biol 4:137–146. Li D, Sham PC, Owen MJ, He L. 2006. Meta-analysis shows significant Petronis A, Van Tol HH, Lichter JB, Livak KJ, Kennedy JL. 1993. The D4 association between dopamine system genes and attention deficit hyper dopamine receptor gene maps on 11p proximal to HRAS. Genomics activity disorder (ADHD). Hum Mol Genet 15:2276–2284. 18:161–163. Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN. 2003. Qian Q, Wang Y, Zhou R, Yang L, Faraone SV. 2004. Family-based and Meta-analysis of genetic association studies supports a contribution of case-control association studies of DRD4 and DAT1 polymorphisms in common variants to susceptibility to common disease. Nat Genet 33: Chinese attention deficit hyperactivity disorder patients suggest long 177–182. repeats contribute to genetic risk for the disorder. Am J Med Genet Part B Lowe N, Kirley A, Mullins C, Fitzgerald M, Gill M, Hawi Z. 2004. Multiple 128B:84–89. marker analysis at the promoter region of the DRD4 gene and ADHD: Raudenbush SW, Bryk AS. 2002. Hierarchical linear models: Applications Evidence of linkage and association with the SNP -616. Am J Med Genet – and data analysis methods. Thousand Oaks: Sage Publications. xxiv, Part B 131B:33 37. 485. pp. Lunetta KL, Faraone SV, Biederman J, Laird NM. 2000. Family-based tests Roman T, Schmitz M, Polanczyk G, Eizirik M, Rohde LA, Hutz MH. 2001. of association and linkage that use unaffected sibs, covariates, and Attention-deficit hyperactivity disorder: A study of association with both – interactions. Am J Hum Genet 66:605 614. the dopamine transporter gene and the dopamine D4 receptor gene. Am – Maher BS, Marazita ML, Ferrell RE, Vanyukov MM. 2002. Dopamine J Med Genet Part B 105B:471 478. system genes and attention deficit hyperactivity disorder: A meta-analy Rowe DC, Stever C, Giedinghagen LN, Gard JM, Cleveland HH, Terris ST, sis. Psychiatr Genet 12:207–215. Mohr JH, Sherman S, Abramowitz A, Waldman ID. 1998. Dopamine DRD4 receptor polymorphism and attention deficit hyperactivity disor Manor I, Tyano S, Eisenberg J, Bachner-Melman R, Kotler M, Ebstein – R. 2002. The short DRD4 repeats confer risk to attention deficit der. Mol Psychiatry 3:419 426. hyperactivity disorder in a family-based design and impair performance Rowe DC, Stever C, Chase D, Sherman S, Abramowitz A, Waldman ID. on a continuous performance test (TOVA). Mol Psychiatry 7:790– 2001. Two dopamine genes related to reports of childhood retrospective 794. inattention and conduct disorder symptoms. Mol Psychiatry 6:429–433. Martinez-Levy G, John DG, Magdalena BV, Ariadna GS, Francisco de LP, Sagvolden T, Johansen EB, Aase H, Russell VA. 2005. A dynamic develop Liz SM, Lino PC, Josefina RG, Ernesto RZ, Carlos CF. 2009. Genetic mental theory of attention-deficit/hyperactivity disorder (ADHD) predominantly hyperactive/impulsive and combined subtypes. Behav Moyzis R, Posner MI. 2000. Attention deficit/hyperactivity disorder Brain Sci 28:397–419. children with a 7-repeat allele of the dopamine receptor D4 gene have extreme behavior but normal performance on critical neuro Seeman P, Van Tol H. 1994. Dopamine receptor pharmacology. Trends in psychological tests of attention. Proc Natl Acad Sci USA 97:4754– pharmacological sciences 15:264–270. 4759. Shaw P, Gornick M, Lerch J, Addington A, Seal J, Greenstein D, Sharp W, Tahir E, Yazgan Y, Cirakoglu B, Ozbay F, Waldman I, Asherson PJ. 2000. Evans A, Giedd JN, Castellanos FX, Rapoport JL. 2007. Polymorphisms of Association and linkage of DRD4 and DRD5 with attention deficit the dopamine D4 receptor, clinical outcome, and cortical structure hyperactivity disorder (ADHD) in a sample of Turkish children. Mol in attention-deficit/hyperactivity disorder. Arch Gen Psychiatry 64: Psychiatry 5:396–404. 921–931. Thapar A, Langley K, O’donovan M, Owen M. 2006. Refining the attention Smalley SL, Bailey JN, Palmer CG, Cantwell DP, McGough JJ, Del’Homme deficit hyperactivity disorder phenotype for molecular genetic studies. MA, Asarnow JR, Woodward JA, Ramsey C, Nelson SF. 1998. Evidence Mol Psychiatry 11:714–720. that the dopamine D4 receptor is a susceptibility gene in attention deficit hyperactivity disorder. Mol Psychiatry 3:427–430. Thompson SG, Higgins JP. 2002. How should meta-regression analyses be undertaken and interpreted? Stat Med 21:1559–1573. Smith KM, Daly M, Fischer M, Yiannoutsos CT, Bauer L, Barkley R, Navia BA. 2003. Association of the dopamine beta hydroxylase gene with Todd RD, Neuman RJ, Lobos EA, Jong YJ, Reich W, Heath AC. 2001. Lack attention deficit hyperactivity disorder: Genetic analysis of the Milwau of association of dopamine D4 receptor gene polymorphisms with kee longitudinal study. Am J Med Genet Part B 119B:77–85. ADHD subtypes in a population sample of twins. Am J Med Genet Part B 105B:432–438. Sunohara GA, Roberts W, Malone M, Schachar RJ, Tannock R, Basile VS, Wigal T, Wigal SB, Schuck S, Moriarty J, Swanson JM, Kennedy JL, Barr Van Tol H, Bunzow J, Guan H, Sunahara R, Seeman P, Niznik H, Civelli O. CL. 2000. Linkage of the dopamine D4 receptor gene and attention 1991. Cloning of the gene for a human dopamine D4 receptor with high deficit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry affinity for the antipsychotic clozapine. Nature 350:614–619. 39:1537–1542. Waldman ID, Rowe DC, Abramowitz A, Kozel ST, Mohr JH, Sherman SL, Swanson JM, Sunohara GA, Kennedy JL, Regino R, Fineberg E, Wigal T, Cleveland HH, Sanders ML, Gard JM, Stever C. 1998. Association and Lerner M, Williams L, LaHoste GJ, Wigal S. 1998. Association of the linkage of the dopamine transporter gene and attention-deficit hyperac dopamine receptor D4 (DRD4) gene with a refined phenotype of tivity disorder in children: Heterogeneity owing to diagnostic subtype attention deficit hyperactivity disorder (ADHD): A family-based ap and severity. Am J Hum Genet 63:1767–1776. proach. Mol Psychiatry 3:38–41. West A, Langley K, Hamshere ML, Kent L, Craddock N, Owen MJ, Swanson J, Oosterlaan J, Murias M, Schuck S, Flodman P, Spence MA, O’Donovan M, Thapar A. 2002. Evidence to suggest biased phenotypes Wasdell M, Ding Y, Chi HC, Smith M, Mann M, Carlson C, Kennedy JL, in children with Attention Deficit Hyperactivity Disorder from Sergeant JA, Leung P, Zhang YP, Sadeh A, Chen C, Whalen CK, Babb KA, completely ascertained trios. Mol Psychiatry 7:962–966.